Photolithographically patterned self-bending spring structures (e.g., spring probes) have been developed, for example, to produce low cost probe cards and to provide electrical connections between integrated circuits. A typical self-bending spring structure is formed from a stress-engineered (a.k.a. “stressy”) metal film that is intentionally fabricated such that its lower/upper portions have a higher internal tensile stress than its upper/lower portions. For example, a spring bending away from a substrate surface has lower tensile stress in the lower portion than in the upper portion, thus producing an upward bend (note that all of the examples provided herein describe this stress gradient). In contrast, a downward bending spring may be produced by providing a higher tensile stress in the lower portion than in the upper portion. The internal stress gradient is produced in the stress-engineered metal film by layering different metals having the desired stress characteristics, or using a single metal by altering the fabrication parameters during deposition. The stress-engineered metal film is patterned to form islands that are secured to an underlying substrate either directly or using an intermediate release material layer. When the release material (and/or underlying substrate) is selectively etched from beneath a first (free) portion, the free portion bends away from the substrate to relieve the internal stress, thereby producing a spring structure that remains secured to the substrate by an anchor portion, but has a bent “free” (cantilevered) portion that extends away from the substrate surface. The tip of the cantilevered portion may then be contacted with selected pads on an integrated circuit, or curvature of the spring structure may be controlled to form a loop or other desired shape. In this manner, such spring structure may be used in probe cards, for electrically bonding integrated circuits, circuit boards, and electrode arrays, and for producing other devices such as inductors, variable capacitors, and actuated mirrors. Examples of such spring structures are disclosed in U.S. Pat. No. 3,842,189 (Southgate) and U.S. Pat. No. 5,613,861 (Smith).
When used to form probe cards, such spring metal structures must exhibit sufficient stiffness to facilitate proper electrical connection between the probe (spring metal finger) and a corresponding contact pad on the device-under-test. Most stressy metal spring probes produced by conventional methods are fabricated from sputtered or plated metal that is approximately one micron thick, which produces only a nominal stiffness capable of resisting a force of 0.1 to 0.2 grams (gmf). These stressy metal spring probes may provide sufficient stiffness to probe gold contact pads, but are not stiff enough to reliably probe aluminum pads. Gold pads can be readily probed with relatively weak spring probes because gold does not form a passivation layer that takes significant force to puncture. However, aluminum pads form a passivation layer that must be punctured by the tip of the spring probe in order to facilitate proper electrical connection. To repeatedly achieve electrical contact to aluminum, which is required for many integrated circuit probe card applications, deflection of the probes within their elastic region should absorb an expected force of at least a few grams.
One method of increasing the stiffness of stressy metal spring structures is to increase its thickness by producing thicker stressy metal films. However, the release height of a spring structure is proportional to its stress gradient divided by the stressy metal film thickness. This means that, by making the stressy metal film thicker, the release height is reduced. Of course, one can compensate for this reduced release height by increasing the stress gradient, but there are practical limits to how much stress can,be induced, and the induced stress often cannot be increased enough to compensate for a very thick stressy metal film. Therefore, the (thin) stressy metal film thickness itself is mostly used to tune for a desired release height.
A more desirable approach to generating spring structures having a higher stiffness is to form and release a relatively thin stressy metal structure, and then thickening the structure using a plating process. Most uses for spring structures today utilize plating (a.k.a., “cladding”) of the released springs for improving various spring characteristics such as electrical conductivity, hardness and wear resistance. Plating a thick metal layer (a few microns) on the stress metal film significantly increases probe stiffness, but could also decrease the maximum deflection. Maximum deflection is determined by the initial lift height and the fracture limit of the structure. Laboratory experiments have shown thick electroplated stiffened springs break or yield when deflected a significant fraction of their initial lift height. Failure typically occurs at the base (anchor portion) of the cantilevers, where plating formed either on the bottom surface of the release spring or spontaneously plated onto the underlying substrate surface forms a wedge that acts as a stress-concentrating fulcrum to pry the base away from the underlying substrate as the structure is deflected, resulting in “delamination” of the spring structure. This is currently a serious issue for the reliability of stressed-metal interconnects. Thermocycling results have shown that the current spring structure is very sensitive to delamination. This wedge limits the maximum force of the resulting spring structure because it limits both the allowed thickness of the plating and the maximum displacement.
Another problem associated with plating conventional spring structures is the formation of “resist-edge” plating that is often undesirably deposited around the springs close to the resist mask that defines the release window. A resist-reflow step (e.g. resist annealing, acetone reflow) is often used to avoid the resist-edge plating, but the reflow step does not always reliably prevent the formation of resist-edge plating, and it is also difficult to implement in production.
Accordingly, what is needed is a cost effective method for fabricating spring probes and other spring structures from self-bending spring materials that are thick (stiff) enough to support, for example, large probing forces, but avoid the delamination associated with conventional plated spring structures. What is also needed is a cost effective method for fabricating probes and other spring structures that reliably prevents the formation of resist-edge plating.